Pith. sign in

REVIEW 2 major objections 5 minor 52 references

Wisdom of the Crowds in Forecasting: Forecast Summarization for Supporting Future Event Prediction

T0 review · 2 major / 5 minor · reviewed 2026-08-08 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This survey organizes the emerging body of work that predicts future events by aggregating forecasts expressed in text, and proposes an eight-component data model to make that aggregation more structured.

desk verdict A useful but under-documented survey of a niche; the data model is a reasonable proposal, and the coverage claims need to be made auditable. read the letter →

arxiv 2502.08205 v1 pith:MJWQ4YNW submitted 2025-02-12 cs.LG cs.CLcs.IR

Anisha Saha , Adam Jatowt This is my paper · ORCID
classification cs.LGcs.CLcs.IR
keywords futureeventpredictionwisdomofthecrowdforecastaggregationsummarizationtemporalinformationextractionlargelanguagemodelsdatamodelsurvey
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper is a survey of computational approaches that predict future events by aggregating forecasts expressed in text by many people, a task it calls Future Event Prediction based on Crowd Wisdom (FEP-CW). It is, to the authors' knowledge, the first survey to organize this area, covering 36 papers, the datasets they rely on, methods for extracting future-related statements from news, tweets, and web pages, and techniques for aggregating them into predictions. The paper's main new contribution is a data model that decomposes a future-related statement into eight components: the predicted event, its attributes, genre, source credibility, conditions, rationale, modality, and timestamp. If accepted, this gives researchers a structured map of a young field and a schema for describing forecasts in a way that could make aggregation more principled than treating each forecast as an atomic unit.

What carries the argument

The central object is the paper's proposed data model of a future-related statement (FRS). It breaks a forecast into eight components: Predicted Future Event, Predicted Attribute of Future Event, Genre of Statement, Credibility of a Source, Conditions, Rationale, Modality, and Timestamp. This decomposition is the paper's novel contribution and is positioned as the machinery that would let aggregation go beyond treating forecasts as atomic units — for example, by grouping forecasts by their conditions or rationale, or weighting by credibility and timestamps. The survey is organized around the FEP-CW pipeline: data acquisition, extraction of future-related sentences (by temporal expressions, keyword markers, morphological patterns, or language models), post-filtering of past 'future' statements, and aggregation via frequency analysis, clustering, ranking, or LLM-based reasoning.

What would settle it

A systematic search of publication databases for surveys published before February 2025 that review 'forecast aggregation', 'crowd forecasting', or 'future event prediction from text' under any name would settle the first-survey claim; finding any such prior survey would refute it. For the data model's novelty, a search for earlier schemas that decompose prediction statements into components such as event, time, condition, rationale, and modality — common in argumentation and information-extraction work — would be the test.

Watch

Extended reading notes

Core claim

The authors claim a two-part discovery. First, no previous survey has examined research that generates future-event predictions by detecting, aggregating, and summarizing individual forecasts expressed in text by multiple users; this survey fills that gap by systematizing 36 works into stages: data collection, extraction of future-related information, aggregation, and final prediction formulation. Second, the paper introduces a data model that decomposes a future-related statement into eight components — predicted future event, predicted attribute of the event (especially its date), genre of the statement, credibility of the source, conditions that must hold, rationale, modality, and timestamp — and argues that using these components, rather than whole statements, enables better aggregation, for instance by weighting forecasts by timeliness or credibility or by grouping forecasts with similar conditions and rationale.

Load-bearing premise

The load-bearing premise is that the field is correctly delimited by the criterion that relevant systems must aggregate future-referring statements expressed in text, excluding time-series and historical-pattern forecasting; if other survey authors draw that boundary differently, or if the undocumented keyword search missed earlier work, the survey's 36-paper coverage and its 'first survey' claim would lose their footing.

Editorial extensions

If this is right

  • The data model implies a concrete research program: automatically extracting the eight components from forecasts would enable aggregation that weights by timestamp, credibility, and modality rather than treating every forecast equally.
  • The survey reports that large language models currently underperform the median human crowd in real-world forecasting, and that supplying background context and careful prompts improves LLM forecasts, pointing to near-term gains from retrieval-augmented generation and human-in-the-loop setups.
  • Datasets in FEP-CW age quickly and carry a leakage risk from LLM training cutoffs, so the field needs temporally mindful evaluation protocols; the paper treats this as a central challenge rather than a minor detail.
  • Most existing aggregation methods are unsupervised clustering or ranking because dedicated annotated datasets are scarce; the survey identifies the creation of such datasets as a priority for progress.
  • Combining crowd-wisdom aggregation with other forecasting approaches such as time-series analysis and causal modeling is explicitly listed as a promising future direction.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The eight-component data model reads like a labeling scheme for a future information-extraction task; one could train a single model to tag all components, giving the field a universal annotation layer across datasets like FORECASTQA, Autocast, and ExpTime.
  • The 'conditions' component implies that forecasts form a directed dependency graph (event A is expected if event B occurs), which could be exploited for conflict resolution and for generating conditional timelines, an extension the paper mentions only briefly.
  • The survey's boundary means its 'first survey' claim applies specifically to text-driven aggregation; an integrated survey that also covers numeric and time-series forecasting would be a natural successor, one the paper itself flags as future integration work.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 5 minor

Summary. This paper presents a survey of computational approaches to Future Event Prediction based on Crowd Wisdom (FEP-CW), covering data sources, extraction of future-related statements, aggregation methods, and LLM-based forecasting. It claims to be the first survey on this topic, reports selecting 36 papers via a keyword search, and proposes a novel data model that decomposes future-related statements into eight components: predicted event, attributes, genre, credibility, conditions, rationale, modality, and timestamp. The paper concludes with a list of future research directions.

Significance. If the survey's coverage is accurate, it offers a useful organizational entry point into a fragmented niche, and the proposed data model is a conceptually clear schema for representing individual forecasts. The paper's strengths are its broad reference list, the explicit differentiation of FEP-CW from time-series and historical-pattern forecasting, and the concrete component breakdown in Section 5. However, the central claims of first-ness and comprehensive coverage are not currently auditable because no search protocol or complete list of included studies is provided, and some included works appear to violate the stated inclusion criterion. The data model is not empirically tested, although the paper openly labels that as future work. These issues are fixable and do not negate the basic utility of the survey for newcomers.

major comments (2)
  1. [Section 1 and Section 7] The claim of selecting 36 relevant papers and being the first survey of FEP-CW is not reproducible from the text. The paper does not report the search protocol: no databases, query strings, search dates, screening rules, or the complete list of the 36 papers. Table 1 lists datasets, not studies. This lack of an auditable methodology prevents readers from verifying coverage and undercuts the novelty claim in Section 7. Please provide a PRISMA-style flow or an appendix with the full list of included papers, DOIs, and the exact search steps.
  2. [Sections 1, 2, and 4] The stated inclusion criterion in Section 1 excludes works that extrapolate historical patterns or use time series, yet the survey includes Radinsky and Horvitz (2013) in Section 2, which mines past news to learn temporal patterns, and temporal-knowledge-graph forecasting works (Deng et al. 2020; Ma et al. 2023a,b; Lee et al. 2023; Zhang and Ning 2024) in Section 4. These works consume historical structured facts rather than future-referring statements issued by multiple users. This inconsistency affects the boundary of the corpus and the 'first survey' claim. The authors should either broaden the inclusion criterion to explicitly cover such text-derived historical evidence and justify their inclusion, or exclude them and update the corpus size accordingly.
minor comments (5)
  1. [Section 2] The paper contains typos: 'Expecially' should be 'Especially' in the paragraph on dataset curation.
  2. [Section 4.4] The text has 'usefil' instead of 'useful' in the sentence 'with LLMs becoming very usefil'.
  3. [Section 4.2] The phrase 'The later post-clustering step' should grammatically be 'The latter post-clustering step'.
  4. [Table 1] The 'Structured' column header is ambiguous; clarify whether it refers to the dataset format, the annotations, or the collection method. Also consider renaming 'Count' to 'Number of Instances' for clarity.
  5. [Section 4.4 and References] There is a missing space in the citation 'GPT-4's[Achiam et al., 2023]', and the reference list contains a duplicate entry for 'Liu, 2019' (one as 'Liu, Yinhan' and one as 'Yinhan Liu'); unify these.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the survey and proposed data model are descriptive, and self-citations serve as surveyed literature rather than as load-bearing premises.

full rationale

This paper is a survey plus a proposed data model; it contains no derivation chain that could reduce to its own inputs. The central claims are literature claims: that 36 relevant papers were selected and that this is the first survey of FEP-CW. Those claims rest on the stated inclusion criterion and on an undocumented keyword search, which raises transparency and coverage concerns, but not circularity: the conclusion is not assumed in the premises. The data model in Section 5 decomposes future-related statements into components such as Predicted Future Event, Conditions, Rationale, and Modality; it is a proposed ontology, not a result derived from the surveyed papers or from the authors' prior work. The paper does cite many works by the same authors, both as examples of the area and as sources of datasets and methods, but those citations are used as evidence that the research area exists and as surveyed content, not as an external theorem or fitted parameter that forces the survey's claims. The skeptical observation that some included works, such as Radinsky and Horvitz (2013) and temporal-knowledge-graph forecasting papers, may violate the stated text-based crowd-forecast criterion is an internal-consistency and reproducibility issue, not a circularity issue. The manuscript also explicitly flags limitations about evaluation and data contamination in Section 7, but no passage asserts that a prediction or first-principles result is equivalent to its inputs by construction. Therefore the appropriate finding is no significant circularity.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

No physical or scientific entities are postulated. The data model is a conceptual taxonomy, not an invented entity with independent falsifiable evidence. There are no fitted free parameters because the paper contains no quantitative experiments.

assumptions (3)
  • domain assumption Works that use future-referring statements in text are the proper scope of FEP-CW; works based on historical patterns or time series are excluded.
    Section 1 defines the survey criterion, which determines the 36-paper corpus and the 'first survey' claim.
  • domain assumption Aggregating many individual forecasts yields more reliable predictions than single forecasts.
    Section 1 states the motivating premise: 'Collective judgment and verdict given by a group of people is often more reliable and accurate compared to that of an individual.'
  • ad hoc to paper The eight proposed data model components are useful for aggregation and can be automatically extracted.
    Section 5 proposes the model but provides no experiments; Section 7 defers extraction experiments to future work.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Wisdom of the Crowds in Forecasting: Forecast Summarization for Supporting Future Event Prediction." pith.science (2026). https://pith.science/paper/MJWQ4YNW

@misc{pith2026250208205,
  author       = {Pith},
  title        = {Pith review of: Wisdom of the Crowds in Forecasting: Forecast Summarization for Supporting Future Event Prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MJWQ4YNW}},
  note         = {Machine review of arXiv:2502.08205}
}
read the original abstract

Future Event Prediction (FEP) is an essential activity whose demand and application range across multiple domains. While traditional methods like simulations, predictive and time-series forecasting have demonstrated promising outcomes, their application in forecasting complex events is not entirely reliable due to the inability of numerical data to accurately capture the semantic information related to events. One forecasting way is to gather and aggregate collective opinions on the future to make predictions as cumulative perspectives carry the potential to help estimating the likelihood of upcoming events. In this work, we organize the existing research and frameworks that aim to support future event prediction based on crowd wisdom through aggregating individual forecasts. We discuss the challenges involved, available datasets, as well as the scope of improvement and future research directions for this task. We also introduce a novel data model to represent individual forecast statements.

Figures

Figures reproduced from arXiv: 2502.08205 by the authors.

Figure 1
Figure 1. General Flow of Predicting Future Event works that perform future forecasting by aggregating multi￾ple future-related statements expressed in text. This means that researches that study patterns of historical events (e.g., frequencies, chronology, causality, etc.) to extrapolate them to the future were not included in our survey, same as works which utilize time series for making the predictions. Our main criterion … view at source ↗
Figure 2
Figure 2. Atomic Components of Individual Future Prediction that [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

52 extracted references · 45 canonical work pages

  1. [1]

    Gpt-4 technical report

    [Achiam et al., 2023] Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Alt- man, Shyamal Anadkat, et al. Gpt-4 technical report. arXiv preprint arXiv:2303.08774,

    arXiv 2023

  2. [10]

    Future retrieval: What does the future talk about? In Workshop on Enriching Information Retrieval of the 34th ACM Annual SIGIR Conference (SIGIR 2011), pages 3–pages,

    [Dias et al., 2011] Ga¨el Dias, Ricardo Campos, and Al´ıpio M ´ario Jorge. Future retrieval: What does the future talk about? In Workshop on Enriching Information Retrieval of the 34th ACM Annual SIGIR Conference (SIGIR 2011), pages 3–pages,

    work page 2011

  3. [11]

    Glm: Gen- eral language model pretraining with autoregressive blank infilling,

    [Du et al., 2022] Zhengxiao Du, Yujie Qian, Xiao Liu, Ming Ding, Jiezhong Qiu, Zhilin Yang, and Jie Tang. Glm: Gen- eral language model pretraining with autoregressive blank infilling,

    work page 2022

  4. [13]

    Bias and fairness in large language models: A survey

    [Gallegos et al., 2024] Isabel O Gallegos, Ryan A Rossi, Joe Barrow, Md Mehrab Tanjim, Sungchul Kim, Franck Der- noncourt, Tong Yu, Ruiyi Zhang, and Nesreen K Ahmed. Bias and fairness in large language models: A survey. Computational Linguistics, pages 1–79,

    work page 2024

  5. [15]

    What happens next? future subevent prediction using contextual hierarchical lstm

    [Hu et al., 2017] Linmei Hu, Juanzi Li, Liqiang Nie, Xiao-Li Li, and Chao Shao. What happens next? future subevent prediction using contextual hierarchical lstm. In Proceed- ings of the AAAI Conference on Artificial Intelligence, vol- ume 31,

    work page 2017

  6. [16]

    Estimating the time between twit- ter messages and future events

    [H¨urriyeto˘glu et al., 2013] Ali H¨urriyeto˘glu, FA Kunneman, and APJ van den Bosch. Estimating the time between twit- ter messages and future events

    work page 2013

  7. [18]

    Temporal ranking of search engine results

    [Jatowt et al., 2005] Adam Jatowt, Yukiko Kawai, and Kat- sumi Tanaka. Temporal ranking of search engine results. In Web Information Systems Engineering–WISE 2005: 6th International Conference on Web Information Systems En- gineering, New York, NY, USA, November 20-22,

    work page 2005

  8. [19]

    Supporting analy- sis of future-related information in news archives and the web

    [Jatowt et al., 2009] Adam Jatowt, Kensuke Kanazawa, Satoshi Oyama, and Katsumi Tanaka. Supporting analy- sis of future-related information in news archives and the web. In Proceedings of the 9th ACM/IEEE-CS joint con- ference on Digital libraries, pages 115–124,

    work page 2009

Show all 52 references
  1. [20]

    Analyzing collective view of future, time- referenced events on the web

    [Jatowt et al., 2010] Adam Jatowt, Hideki Kawai, Kensuke Kanazawa, Katsumi Tanaka, Kazuo Kunieda, and Keiji Yamada. Analyzing collective view of future, time- referenced events on the web. In Proceedings of the 19th international conference on World wide web, pages 1123– 1124,

    2010

  2. [22]

    Mapping temporal hori- zons: Analysis of collective future and past related atten- tion in twitter

    [Jatowt et al., 2015] Adam Jatowt, ´Emilien Antoine, Yukiko Kawai, and Toyokazu Akiyama. Mapping temporal hori- zons: Analysis of collective future and past related atten- tion in twitter. In Proceedings of the 24th international conference on World Wide Web, pages 484–494,

    2015

  3. [24]

    Improving retrieval of future-related information in text collections

    [Kanazawa et al., 2011] Kensuke Kanazawa, Adam Jatowt, and Katsumi Tanaka. Improving retrieval of future-related information in text collections. In 2011 IEEE/WIC/ACM International Conferences on Web Intelligence and Intelli- gent Agent Technology, volume 1, pages 278–283. IEEE,

    2011

  4. [25]

    Ranking related news predictions

    [Kanhabua et al., 2011] Nattiya Kanhabua, Roi Blanco, and Michael Matthews. Ranking related news predictions. In Proceedings of SIGIR, pages 755–764,

    2011

  5. [26]

    Chronoseeker: Search engine for future and past events

    [Kawai et al., 2010] Hideki Kawai, Adam Jatowt, Kat- sumi Tanaka, Kazuo Kunieda, and Keiji Yamada. Chronoseeker: Search engine for future and past events. In Proceedings of the 4th International Conference on Uniquitous Information Management and Communica- tion, pages 1–10,

    2010

  6. [27]

    Temporal knowledge graph forecasting without knowledge using in-context learning

    [Lee et al., 2023] Dong-Ho Lee, Kian Ahrabian, Woojeong Jin, Fred Morstatter, and Jay Pujara. Temporal knowledge graph forecasting without knowledge using in-context learning. arXiv preprint arXiv:2305.10613,

    2023 arXiv

  7. [28]

    Morphology and lexical semantics

    [Levin and Hovav, 2017] Beth Levin and Malka Rappaport Hovav. Morphology and lexical semantics. The handbook of morphology, pages 248–271,

    2017

  8. [29]

    Roberta: A robustly optimized bert pretraining approach,

    [Liu et al., 2019] Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. Roberta: A robustly optimized bert pretraining approach,

    2019

  9. [30]

    Roberta: A robustly optimized bert pretraining approach

    [Liu, 2019] Yinhan Liu. Roberta: A robustly optimized bert pretraining approach. arXiv preprint arXiv:1907.11692 , 364,

    2019 arXiv

  10. [31]

    Structured, complex and time-complete temporal event forecasting

    [Ma et al., 2023b] Yunshan Ma, Chenchen Ye, Zijian Wu, Xiang Wang, Yixin Cao, Liang Pang, and Tat-Seng Chua. Structured, complex and time-complete temporal event forecasting. arXiv preprint arXiv:2312.01052,

    arXiv

  11. [32]

    Robust temporal processing of news

    [Mani and Wilson, 2000] Inderjeet Mani and George Wil- son. Robust temporal processing of news. In Proceedings of the 38th annual meeting of the association for compu- tational linguistics, pages 69–76,

    2000

  12. [34]

    Monte carlo simu- lation

    [Mooney, 1997] Christopher Z Mooney. Monte carlo simu- lation. Number

    1997

  13. [38]

    Automatic extraction of references to future events from news articles using se- mantic and morphological information

    [Nakajima, 2015] Yoko Nakajima. Automatic extraction of references to future events from news articles using se- mantic and morphological information. In Proceedings of the 24th International Conference on Artificial Intelli- gence, IJCAI’15, page 4385–4386. AAAI Press,

    2015

  14. [39]

    Navigating tomorrow: Reliably assessing large language models performance on future event prediction,

    [Nako and Jatowt, 2025] Petraq Nako and Adam Jatowt. Navigating tomorrow: Reliably assessing large language models performance on future event prediction,

    2025

  15. [40]

    Computational exploration to linguistic structures of future: Classification and categorization

    [Ni et al., 2015] Aiming Ni, Jinho D Choi, Jason Shepard, and Phillip Wolff. Computational exploration to linguistic structures of future: Classification and categorization. In Proceedings of NAACL: Student research workshop, pages 168–173,

    2015

  16. [41]

    Training language models to follow instruc- tions with human feedback

    [Ouyang et al., 2022] Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. Training language models to follow instruc- tions with human feedback. Advances in neural informa- tion p...

    2022

  17. [42]

    Evaluating encoder models for future event forecasting/submitted by fabio pernegger

    [Pernegger, 2024] Fabio Pernegger. Evaluating encoder models for future event forecasting/submitted by fabio pernegger

    2024

  18. [43]

    Mining the web to predict future events

    [Radinsky and Horvitz, 2013] Kira Radinsky and Eric Horvitz. Mining the web to predict future events. In Proceedings of the sixth ACM international conference on Web search and data mining, pages 255–264,

    2013

  19. [44]

    A survey on bias de- tection in online news using deep learning

    [Rakhecha et al., 2023] Khushi Rakhecha, Simran Rauniyar, Muskan Agrawal, and Aruna Bhatt. A survey on bias de- tection in online news using deep learning. In 2023 2nd International Conference on Applied Artificial Intelligence and Computing (ICAAIC), pages 396–403,

    2023

  20. [45]

    Future timelines: Extraction and vi- sualization of future-related content from news articles

    [Regev et al., 2024] Juwal Regev, Adam Jatowt, and Michael F ¨arber. Future timelines: Extraction and vi- sualization of future-related content from news articles. Horizons, 51(62):84,

    2024

  21. [46]

    Large language model prediction capabilities: Evidence from a real-world forecasting tournament

    [Schoenegger and Park, 2023] Philipp Schoenegger and Pe- ter S Park. Large language model prediction capabilities: Evidence from a real-world forecasting tournament. arXiv preprint arXiv:2310.13014,

    2023 arXiv

  22. [47]

    Extraction and exploration of spatio- temporal information in documents

    [Str¨otgen et al., 2010] Jannik Str ¨otgen, Michael Gertz, and Pavel Popov. Extraction and exploration of spatio- temporal information in documents. In Proceedings of the 6th Workshop on Geographic Information Retrieval, pages 1–8,

    2010

  23. [48]

    Predicting time-to-event from twitter mes- sages

    [Tops et al., 2013] Hannah Tops, APJ van den Bosch, and FA Kunneman. Predicting time-to-event from twitter mes- sages

    2013

  24. [49]

    Multi- timeline summarization (mtls): Improving timeline sum- marization by generating multiple summaries

    [Yu et al., 2021] Yi Yu, Adam Jatowt, Antoine Doucet, Kazunari Sugiyama, and Masatoshi Yoshikawa. Multi- timeline summarization (mtls): Improving timeline sum- marization by generating multiple summaries. In Pro- ceedings of the 59th Annual Meeting of the Association for Compu...

    2021

  25. [51]

    Large language models as event forecasters,

    [Zhang and Ning, 2024] Libo Zhang and Yue Ning. Large language models as event forecasters,

    2024

  26. [52]

    Forecast- ing future world events with neural networks

    [Zou et al., 2022] Andy Zou, Tristan Xiao, Ryan Jia, Joe Kwon, Mantas Mazeika, Richard Li, Dawn Song, Jacob Steinhardt, Owain Evans, and Dan Hendrycks. Forecast- ing future world events with neural networks. Advances in Neural Information Processing Systems, 35:27293–27305, 2022

    2022

  27. [1997]

    Analyzing the role of context in fore- casting with large language models

    [Mutschlechner and Jatowt, 2025] Gerrit Mutschlechner and Adam Jatowt. Analyzing the role of context in fore- casting with large language models. arXiv preprint arXiv:2501.06496,

    2025 arXiv

  28. [2000]

    The language of time: a reader

    [Mani et al., 2005] Inderjeet Mani, James Pustejovsky, and Robert Gaizauskas. The language of time: a reader . Ox- ford University Press, USA,

    2005

  29. [2005]

    What is the temporal value of web snippets?

    [Campos et al., 2011] Ricardo Campos, Al ´ıpio Jorge, and Ga¨el Dias. What is the temporal value of web snippets?

    2011

  30. [2009]

    Searching the fu- ture

    [Baeza-Yates, 2005] Ricardo Baeza-Yates. Searching the fu- ture. In SIGIR Workshop MF/IR, volume 5,

    2005

  31. [2010]

    Multi-lingual analysis of future-related informa- tion on the web

    [Jatowt et al., 2013] Adam Jatowt, Hideki Kawai, Kensuke Kanazawa, Katsumi Tanaka, Kazuo Kunieda, and Keiji Yamada. Multi-lingual analysis of future-related informa- tion on the web. In 2013 International Conference on Cul- ture and Computing, pages 27–32. IEEE,

    2013

  32. [2011]

    Identifying top relevant dates for implicit time sensitive queries

    [Campos et al., 2017] Ricardo Campos, Ga ¨el Dias, Al´ıpio M ´ario Jorge, and C ´elia Nunes. Identifying top relevant dates for implicit time sensitive queries. Information Retrieval Journal, 20:363–398,

    2017

  33. [2013]

    Extracting collective expectations about the fu- ture from large text collections

    [Jatowt and Au Yeung, 2011] Adam Jatowt and Ching-man Au Yeung. Extracting collective expectations about the fu- ture from large text collections. In Proceedings of the 20th ACM international conference on Information and knowl- edge management, pages 1259–1264,

    2011

  34. [2014]

    Future predic- tion with automatically extracted morphosemantic pat- terns

    [Nakajima et al., 2020] Yoko Nakajima, Michal Ptaszynski, Fumito Masui, and Hirotoshi Honma. Future predic- tion with automatically extracted morphosemantic pat- terns. Cognitive Systems Research, 59:37–62,

    2020

  35. [2015]

    Forecastqa: a question answering challenge for event forecasting with temporal text data

    [Jin et al., 2020] Woojeong Jin, Rahul Khanna, Suji Kim, Dong-Ho Lee, Fred Morstatter, Aram Galstyan, and Xi- ang Ren. Forecastqa: a question answering challenge for event forecasting with temporal text data. arXiv preprint arXiv:2005.00792,

    2020 arXiv

  36. [2017]

    Learning towards abstractive timeline summarization

    [Chen et al., 2019] Xiuying Chen, Zhangming Chan, Shen Gao, Meng-Hsuan Yu, Dongyan Zhao, and Rui Yan. Learning towards abstractive timeline summarization. In IJCAI, pages 4939–4945,

    2019

  37. [2018]

    Clustering and exploring search results using timeline constructions

    [Alonso et al., 2009] Omar Alonso, Michael Gertz, and Ri- cardo Baeza-Yates. Clustering and exploring search results using timeline constructions. In David Wai-Lok Cheung, Il-Yeol Song, Wesley W. Chu, Xiaohua Hu, and Jimmy Lin, editors, Proceedings of the 18th ACM Conference o...

    2009

  38. [2019]

    Dynamic knowledge graph based multi- event forecasting

    [Deng et al., 2020] Songgaojun Deng, Huzefa Rangwala, and Yue Ning. Dynamic knowledge graph based multi- event forecasting. In Proceedings of SIGKDD , pages 1585–1595,

    2020

  39. [2020]

    Bert: Pre-training of deep bidirectional transformers for language understand- ing,

    [Devlin et al., 2019] Jacob Devlin, Ming-Wei Chang, Ken- ton Lee, and Kristina Toutanova. Bert: Pre-training of deep bidirectional transformers for language understand- ing,

    2019

  40. [2021]

    Back to the future: To- wards explainable temporal reasoning with large language models

    [Yuan et al., 2023] Chenhan Yuan, Qianqian Xie, Jimin Huang, and Sophia Ananiadou. Back to the future: To- wards explainable temporal reasoning with large language models. arXiv preprint arXiv:2310.01074,

    2023 arXiv

  41. [2022]

    A multidimensional dataset based on crowdsourcing for analyzing and detecting news bias

    [F¨arber et al., 2020] Michael F ¨arber, Victoria Burkard, Adam Jatowt, and Sora Lim. A multidimensional dataset based on crowdsourcing for analyzing and detecting news bias. In Proceedings of the 29th ACM international conference on information & knowledge management , pages ...

    2020

  42. [2023]

    Automatic identification of arabic expressions re- lated to future events in lebanon’s economy.arXiv preprint arXiv:1805.11603,

    [Al-Hajj and Sabra, 2018] Moustafa Al-Hajj and Amani Sabra. Automatic identification of arabic expressions re- lated to future events in lebanon’s economy.arXiv preprint arXiv:1805.11603,

    2018 arXiv

  43. [2024]

    Embedding time expressions for deep temporal ordering models

    [Goyal and Durrett, 2019] Tanya Goyal and Greg Durrett. Embedding time expressions for deep temporal ordering models. arXiv preprint arXiv:1906.08287,

    2019 arXiv

  44. [2025]

    Investigation of fu- ture reference expressions in trend information

    [Nakajima et al., 2014] Yoko Nakajima, Michal Ptaszynski, Hirotoshi Honma, and Fumito Masui. Investigation of fu- ture reference expressions in trend information. In 2014 AAAI Spring Symposium Series,

    2014

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.